8 New Trend in Instrumentation of NIR Spectroscopy …
207
Table 8.2 Millet analysis toward gallic acid equivalents (GAE) by benchtop and ultra-portable
NIR spectrometers
Spectrometer Sample form R 2 (CV) RMSECV/mgGAE/g R 2 (TV) RMSEP/mgGAE/g
NIRFlex
N-500
Intact
0.953
0.365
0.940
0.467
Milled
0.985
0.223
0.920
0.479
SCiO
Intact
0.876
0.601
0.814
0.806
Milled
0.8240
0.743
0.782
0.840
RMSECV and RMSEP values resulting from PLS-R models for protein content (7–14% w/w) are
used here as the parameters describing the prediction performance
CV cross-validated regressions; TV test set-validated regressions
Table 8.3 Cheese analysis toward gallic acid equivalents (GAE) by benchtop and ultra-portable
NIR spectrometers
Spectrometer Sample form R 2 (CV) RMSECV/mgGAE/g R 2 (TV) RMSEP/mgGAE/g
NIRFlex
N-500
Intact
0.9726
1.5711
0.9431
1.8964
Grated
0.9930
0.7845
0.9913
0.7676
SCiO
Intact
0.9801
1.2466
0.9838
1.1874
Grated
0.9838
1.0527
0.9940
0.8194
RMSECV and RMSEP values resulting from PLS-R models for protein content (9–36% w/w) are
used here as the parameters describing the prediction performance
CV cross-validated regressions; TV test set-validated regressions
Nonetheless, their potential is demonstrated even for challenging analysis of multiconstituent natural products where content variability and matrix effects are comparably stronger than most other types of samples. The discussed problem of performance in certain scenario is essential in various fields of research and analysis. Often,
practical applications benefit largely from routine measurements being carried out
on portable instrumentation, however, with calibration and validation controlled via
benchtop spectrometer as the reference. Transfer of multivariate models between
these devices is an essential feature and remains a focused lane of research. For
instance, transferability of spectra and the resulting qualitative and quantitative
calibrations, between benchtop and miniaturized spectrometers was explored by
Hoffman et al. [20]. Recent literature demonstrates keen interest in the performance
of handheld spectrometers in a variety of applications, e.g., material identification and
product authentication [21]. Applicability and performance in quantitative analysis
of miniaturized NIR spectrometers were also conducted in the context of pharmaceutical formulation [22]. Because of on-site capacity, miniaturized NIR spectrometers
have very high potential for applications in the analysis of natural products. However,
complex composition and demanding conditions in such analyses are challenging
for accurate quantitative analysis. Wiedemair and Huck evaluated the performance
of three different miniaturized NIR spectrometers in determining the antioxidant
capacity of gluten-free grains [23].
207
Table 8.2 Millet analysis toward gallic acid equivalents (GAE) by benchtop and ultra-portable
NIR spectrometers
Spectrometer Sample form R 2 (CV) RMSECV/mgGAE/g R 2 (TV) RMSEP/mgGAE/g
NIRFlex
N-500
Intact
0.953
0.365
0.940
0.467
Milled
0.985
0.223
0.920
0.479
SCiO
Intact
0.876
0.601
0.814
0.806
Milled
0.8240
0.743
0.782
0.840
RMSECV and RMSEP values resulting from PLS-R models for protein content (7–14% w/w) are
used here as the parameters describing the prediction performance
CV cross-validated regressions; TV test set-validated regressions
Table 8.3 Cheese analysis toward gallic acid equivalents (GAE) by benchtop and ultra-portable
NIR spectrometers
Spectrometer Sample form R 2 (CV) RMSECV/mgGAE/g R 2 (TV) RMSEP/mgGAE/g
NIRFlex
N-500
Intact
0.9726
1.5711
0.9431
1.8964
Grated
0.9930
0.7845
0.9913
0.7676
SCiO
Intact
0.9801
1.2466
0.9838
1.1874
Grated
0.9838
1.0527
0.9940
0.8194
RMSECV and RMSEP values resulting from PLS-R models for protein content (9–36% w/w) are
used here as the parameters describing the prediction performance
CV cross-validated regressions; TV test set-validated regressions
Nonetheless, their potential is demonstrated even for challenging analysis of multiconstituent natural products where content variability and matrix effects are comparably stronger than most other types of samples. The discussed problem of performance in certain scenario is essential in various fields of research and analysis. Often,
practical applications benefit largely from routine measurements being carried out
on portable instrumentation, however, with calibration and validation controlled via
benchtop spectrometer as the reference. Transfer of multivariate models between
these devices is an essential feature and remains a focused lane of research. For
instance, transferability of spectra and the resulting qualitative and quantitative
calibrations, between benchtop and miniaturized spectrometers was explored by
Hoffman et al. [20]. Recent literature demonstrates keen interest in the performance
of handheld spectrometers in a variety of applications, e.g., material identification and
product authentication [21]. Applicability and performance in quantitative analysis
of miniaturized NIR spectrometers were also conducted in the context of pharmaceutical formulation [22]. Because of on-site capacity, miniaturized NIR spectrometers
have very high potential for applications in the analysis of natural products. However,
complex composition and demanding conditions in such analyses are challenging
for accurate quantitative analysis. Wiedemair and Huck evaluated the performance
of three different miniaturized NIR spectrometers in determining the antioxidant
capacity of gluten-free grains [23].